Can increased support to foster care families reduce the number of moves for children in out-of-home care? Evidence from Norway

Norwegian youth in out-of-home care move three times as frequently as their peers. Such placement instability is linked to negative outcomes in terms of social attachment, well-being, educational achievements, health, and future opportunities. Norway implemented a new child welfare service reform in 2022 that increased the municipalities responsibilities for out-of-home care. The “ incentive package ” and “ Barneløftet ” were measures implemented to prepare the municipalities for these changes. This study evaluates how the implemented measures affect the number of moves within out-of-home care in Trøndelag county. An event-study design with difference-in-difference estimates was used to study the effect of the measures. The data are Norwegian registers that include most children in out-of-home care from 2013 to 2021. The findings of this study indicate that increased support for foster care homes significantly reduces the number of moves. Increased placement stability is associated with an increased sense of belonging, thus facilitating positive development.


INTRODUCTION
The number of moves within child welfare services is a constant concern because the implications that this has on youths (Paulsen & Ytreland, 2022).Instability in placement for youths in out-of-home care (OOHC) is associated with challenges related to social attachment, belonging, reduced well-being, decreased opportunities in adult life, lower educational achievements, and physical and mental development (Backe-Hansen et al., 2014;Bengtsson & Mølholt, 2018;Kääriälä, 2020;Konijn et al., 2019;Paulsen & Ytreland, 2022;Sebba & Luke, 2019;Vinnerljung et al., 2017).Preventing placement instability is thus an explicit goal for policymakers when developing Norwegian child welfare services, and different incentives and reforms are initiated to reach such goals (NOU 2018(NOU :18, 2018;;Paulsen et al., 2022; The Norwegian Association of Local and Regional Authorities, 2021).
A national child welfare services reform was implemented in January 2022 that had major impacts on the municipalities, with increased responsibility.They were now the main responsible for OOHC within their municipality.In the years leading up to this reform, Trøndelag county in central Norway implemented several incentives to meet the criteria in the new reform.A central objective with these incentives was to decrease the number of moves in OOHC in the region by increasing and standardizing supervision and support to foster care families.
This article is concerned with evaluating if these incentives have affected the number of placements for youths in out-of-home care in Trøndelag county in the years following these incentives.
Here, we used an event-study design with differencein-difference estimates to study the effect from the abovementioned measures on the number of moves for children in OOHC.Our data are from Norwegian registers on most children in OOHC from 2013 to 2021.The available measure on moves in the registers is the number of moves between municipalities in a year.This means that we do not measure all moves for children in OOHC-we focus on those who imply a longer geographical move, which usually implies a change in school and social network.Further, the measure does not allow for a distinction between planned moves and unplanned moves-all relocations will therefore be included in the analysis.
With respect to the number of moves between municipalities, children under the age of 18 move on average of 0.03 times per year in the period.For children in OOHC, the average number of moves is over 0.09 (Table 1).This implies that roughly 3% of children move every year in the total population, in contrast to 9% of children in OOHC.This means that Norwegian children in OOHC move three times as frequently as other children (and more than five if we include initial placement).This has important implications.This study analyzes how the implemented measures can contribute to reducing the number of moves.Our findings suggest that increased support for foster care homes significantly reduces the number of moves, although the effect seems to be shortterm.The positive effect on placement stability is believed to strengthen children's sense of belonging.

THEORY A sense of belonging
A sense of belonging is a fundamental human need (Allen et al., 2021).Belonging can be defined as being part of surrounding systems such as communities, physical spaces, social contexts such as family and friends, and cultural groups.Belonging is not something definite, but an ongoing dynamic process that is often an unconscious part of our daily life (Allen et al., 2021;Bengtsson & Luckow, 2020).Belonging can be traced back to humans' biology where the need to be part of social networks and spatial places is related to identity and safety and thus connected to physical health and safety while also being a social matter (Allen et al., 2021;Nethercote, 2022).Belonging is thus necessary for stability and security, which in turn creates a sense of comfort.Not feeling belonging can result in negative feelings such as insecurity, fear, or social rejection (Anderson, 2021;Massey, 1992).
A sense of belonging is often linked to a place, such as the home, as a source of rootedness.This is where social and material processes are bound together through emotional and social meanings.However, developing belonging to people and places requires time.For most people, a sense of belonging is created unconsciously through everyday activities by simply being in the environment in which you create belonging (Bengtsson & Luckow, 2020;Bengtsson & Mølholt, 2018;Bessell, 2023;Christensen et al., 2000;Wade, 2019).While many experience belonging to one place called "home," this is not the case for everyone.Some feel belonging to several social constructions and places-these people might experience challenges related to adapting between different senses of belonging.If these changes between settings of belonging occur frequently, then the individual might have a more conscious relationship to the process of attaching to different material and non-material elements such as space, people, emotions, and other functional attachments.In turn, this person might fail to accomplish belonging and is exposed to feelings such as insecurity and social rejection (Bengtsson & Luckow, 2020).
Being removed involuntarily from a place where you feel belonging can be described as displacement (Elliott-Cooper et al., 2020).Displacement has also been used to describe placement instability among children and youth in foster care (Gauthier et al., 2004;Golden Guzman, 2023 Richardson et al., 2018).When an individual is displaced from their known place it might affect them socially, financially, and emotionally.It is thus also a psychological phenomenon.Displacement has psychological impacts on the individuals because of the emotional or affective separation from a place where an individual feels belonging.There are also psychological impacts because of disruptions of social relationships such as primary caregivers, siblings, or friends (Atkinson, 2015;Baeten et al., 2017;Konijn et al., 2019;Oosterman et al., 2007;Richardson et al., 2018;Westin, 2021).Forced mobility can be motivated by the social good to give the individual better opportunities in another spatial context that has some characteristics that are thought to help the individual; nevertheless, the individual can experience this as being forcefully removed from a community or home where the individual felt belonging (Baeten et al., 2017;Elliott-Cooper et al., 2020).

Placement instability for youths in OOHC
The length of a foster care stay can vary both in terms of length and aim.However, one of the main objectives is to give the child a sense of permanency and stability and allow them to create meaningful relationships.This stability is emphasized as necessary for the child's positive development (Kääriälä, 2020;Konijn et al., 2019;NOU 2018NOU :18, 2018)).However, many youth in foster care experience placement instability and turbulence when they enter OOHC-stays are short, they have to relate to multiple social workers, and they end up with a fragmented experience of belonging (Andersson, 2009;Backe-Hansen et al., 2019;Kääriälä, 2020;Konijn et al., 2019;Mendes & Moslehuddin, 2004;Valentine, 2003).
Placement instability can occur for several reasons such as difficulties related to the biological family, the foster family, the youth themselves, or with child welfare services.Relocation in OOHC is often complex and consist of several factors; however, child-focused factors are often emphasized as the primary cause (Oosterman et al., 2007;Paulsen & Ytreland, 2022;Strijker et al., 2011;Vanderwill et al., 2021;Vinnerljung et al., 2017).Characteristics in the youths behavior, such as mental health, violence and substance abuse are of importance when youths relocate.However, characteristics in the foster families are also known to affect placements, such as the relationship with other family members, experienced stress, fear, and lack of commitment.Additionally, the dynamics with the biological family can be influential.Furthermore, systematic factors such as limited availability of foster families and urgency in the placement process impact the success of placements (Backe-Hansen et al., 2013, 2019).
Challenges leading to relocation can accumulate over time and increase their risk of further relocation in the youths' next placement (Konijn et al., 2019).The average age upon entry into OOHC is $5years, and chances of placement breakdowns tend to increase as the youths get older (Backe-Hansen et al., 2013;NOU 2018NOU :18, 2018)).Importantly, it should be noted that relocation is not necessarily caused by conflicts or struggles, it can also stem from family reunification or as scheduled progress in the youths OOHC program.However, it is acknowledged that OOHC placements often occur abruptly, making registration challenging (Backe-Hansen et al., 2013).Thus, we do not have any official statistics on how many experience involuntary relocation but there are indicators of high numbers of such placements (NOU 2018(NOU :18, 2018)).
Placement instability is related to negative outcomes for several reasons (Backe-Hansen et al., 2014;Vinnerljung et al., 2017).One being that the youths lose significant places that are meaningful to them.Moreover, repeating moves can also imply loss of significant relationships if the youths move from their biological family or change caregivers.As a result, they might develop issues with building relationships, attaching to people, and trusting people in the process of reestablishing relationships in their new location (Andersson, 2009;Konijn et al., 2019;Paulsen & Ytreland, 2022).Consequently, these youths might experience a lower quality of life, which is associated with decreased physical and mental development.This can further lead to lower educational achievement, difficulties in the labor market, and financial hardship (Johansson et al., 2023;Kääriälä, 2020;Paulsen et al., 2022).As a result, youth who have experienced placement instability in their childhood often encounter fewer opportunities in their adult life than their peers (Mendes & Moslehuddin, 2004;Paulsen et al., 2022).

The Norwegian context
Child welfare services in Nordic countries are familyoriented.The child welfare system aims to cooperate with families, and voluntary interventions are the norm.Thus, the use of OOHC is considered a last resort after several other measures have been tried.There is an increased use of OOHC in Norway today, even though it is considered a highly invasive measure (Kääriälä, 2020;NOU 2018NOU :18, 2018)).Socio-economic factors are known to be a determiner for selection to child welfare services.OOHC is one measure where this becomes evident due to the high amount of youths from families with high socioeconomic background compared with the presence of the same group in other child welfare services measures, such as those targeting parental skills or child development.Arguably, there seem to be a higher threshold for this group to be involved in child welfare services, thus measures for this group are likely to be more invasive (Kojan & Storhaug, 2021), between 20% and 50% of youths placed in OOHC in Norway experience placement disruption (NOU 2018(NOU :18, 2018;;Paulsen & Ytreland, 2022).To prevent placement disruption, standardized, and systematic supervision is important and especially so at the beginning of a new placement (Backe-Hansen et al., 2019;Paulsen et al., 2022).Consequently, one of the main objectives when developing OOHC in Norway is to secure predictability, safety, and stability to prevent placement instability (NOU 2018(NOU :18, 2018; The Norwegian Association of Local and Regional Authorities, 2021).

Preparing Trøndelag county for the 2022 child welfare services reform
The new reform for child welfare services in Norway was first implemented in January 2022.This had considerable implications for the municipalities because they were given increased responsibilities on several areas that used to be managed by, or in collaboration with, the Office for Children, Youth, and Family Affairs (Bufetat).OOHC services were greatly affected by the new reform because municipalities were given the main responsibility of these services, which includes financing, supervising, and follow-up of youths and families in OOHC.The municipalities are required to take a more holistic approach to youths in foster care to prevent placement instability and provide suitable support (Pedersen et al., 2022).Trøndelag is a county in the central part of Norway that consists of 38 municipalities.Several municipalities take part in inter-municipality collaborations because of differences in municipality size and the number of inhabitants.This results in 22 units in child welfare services for the 38 municipalities.These units are part of Bufetat Central Norway (Bufetat, region Midt-Norge) together with units from Møre & Romsdal county (2021).
Several measures were implemented in the years leading up to the new reform for child welfare services to invest in OOHC specifically and child welfare services in general (The Norwegian Association of Local and Regional Authorities, 2021).The measures aimed at increasing stability and safety.The OOHC should be more holistically focused, and foster parents should experience better training and supervision within the Trøndelag region (The Norwegian Association of Local and Regional Authorities, 2021).The "incentive package for OOHC" was established as a collaboration between Trøndelag county and Bufetat Central Norway and aimed at reducing the number of moves by providing closer supervision for foster home families.The participants in the incentive package were assigned a consistent supervisor to provide safety and stability to the family.This supervisor evaluated the progress and changes in the foster home, offered counseling for the family, and aimed at increasing the foster parents competence simultaneously as the youth were given the opportunity to participate and be involved.The incentive package also offered couples counseling for foster families along with group counseling with other foster families to share experiences.Additionally, those at highest risk of relocation were offered financial compensation to enable them from abstain from employment and increase the capacity for supervision and guidance of the youths.This project started in 2018 and was finalized in its original form in 2020 (Paulsen & Ytreland, 2022).
In January 2020, another incentive to prepare for the new reform was initiated, that is, "Barneløftet" ("BL").Trøndelag county received finances from The Norwegian Directorate for Children, Youth, and Family Affairs (Bufdir) to increase financial and personnel resources in OOHC.A foster home network was established and made available to all actors in OOHC in the region.This network aimed at increasing predictability and equality in the care offered by the municipalities (The Norwegian Association of Local and Regional Authorities, 2021).The foster home network resulted in a document called "package for supervision" (Omsorgspakken) that is a guide for supervision in three parts: one part each for the youths, the parents, and the foster parents.The goal of this package was to give these youths a better, safer, and more predictable life.Another stated goal was that it should prevent unnecessary youth movement (The Norwegian Association of Local and Regional Authorities, 2021).

METHODS AND DATA
We employed rich administrative Norwegian registers on all individuals under the age of 18 including all individuals in foster care and placed under the care (omsorgstiltak) of the child welfare services.The status of foster care is proxied by selecting individuals under the care of child welfare services (omsorgstiltak) living in private households.The numbers correspond with official numbers and includes most children living in foster care in Norway (Bufdir, 2021).The caveat is that it does not include all individuals in foster care.It leaves out the $1300 children each year living in foster homes under other legislative decisions (e.g., emergency homes) by the child welfare services.
To evaluate the impact of the measures introduced, we estimate here a difference-in-difference model by comparing Trøndelag (treated) to counties that did not apply measures to support foster care during the same period (not treated).We exclude Møre & Romsdal and Nordland counties that border Trøndelag to avoid potential contamination effects.To estimate treatment effects and to verify the assumption of parallel trends before treatment, we used a quasi-experimental event-study approach and computed coefficient for leads and lags in treatment as described by Cunningham (2021).The period of study is from 2013 to 2021 where treatment starts in 2018 with 3 years estimating the average treatment effects.
Our identification strategy is that Trøndelag introduced most of the measures in the period as an anticipatory measure pending the forthcoming reform.The reform is imposed by the central government and not by the county themselves.Although, the choice of implementation in Trøndelag as opposed to other counties is not random but are still suitable for an empirical study.In addition, we check for some treatment assignments and do not see any indication that the frequency of moving or the number of children in foster care were different in Trøndelag versus the rest of the country (Table 1).

Data and descriptive statistics
Data are retrieved from Statistics Norway (SSB) and the Norwegian Centre for Research Data (SIKT) through microdata.no(NSD/SSB, 2019).The data contains detailed information from administrative registers about all individuals under the age of 18 in Norway from 2013 to 2021.The panel for the models is based on 14,201 individuals living in foster care during the period.As individuals move in and out of foster care, the panel is unbalanced with waves including between 6500 and 7800 individuals (Tables A1-A3).
The dependent variable is the number of times an individual in OOHC moved to another municipality within a year.We subtract moves that are due to initial placement.We use cross-municipal moves, as this is the only available information on moving frequency from the registers.Thus, we do not count all changes of residence, only those who are between municipalities.Official numbers on moves for children in OOHC are not well documented.Although Beyrer and Dyrhaug (2021) report that in 2020, 74% of children below the age of five, living in OOHC experienced at least one move after initial placement.In our data, 52% of the same children experience a cross municipal move after initial placement.This example indicates that cross municipal moves constitute about two-thirds of moves for children in OOHC.
Individuals in the sample could move between 355 municipalities in 2017, but this decreased to 289 after a reform in 2020.This would naturally contribute to a decrease in the registered number of movements in the period.In Trøndelag, there are 38 municipalities after the reform with 48 before the reform.This implies a reduction of roughly 20%.The counterfactual group had 307 before and 251 after, which is approximately the same reduction of municipalities in both groups, thus making them comparable throughout the period.Table 1 shows the average number of moves for individuals in and outside of foster care.The variation is from 0 and up to 4 moves in a year.
Table 1 shows descriptive statistics for the number of moves in a year for the full population of individuals under the age of 18 both in the treated county (Trøndelag) and the rest of the country.There were no differences in the baseline population with respect to this.The table shows that the average number of moves for individuals in foster care is more than three times as high as for children outside foster care (more than five times if we include the move initiated by placement).The table also shows that most children do not move every year.
Table 2 shows the share of the youth population in foster care in the treated and untreated areas.The table shows that Trøndelag has roughly the same share of youth in foster care than most of the country.

Analytical approach
The empirical analysis in this study applies an eventstudy design to evaluate the causal effects of introducing measures to support foster care families in terms of the frequency of cross-municipal moving of children in foster care.A difference-in-differences estimator with leads and lags as shown in Cunningham (2021)  some crucial assumptions to be satisfied for an eventstudy approach to yield consistent estimates.The first and most important one is the assumption of parallel trends in baseline outcomes, which will be tested by investigating pre-treatment differences (or pre-trends) as measured by the t-statistics for leads.The pre-trends tell us whether there are differences in outcomes between the treated and untreated sample before treatment.
A second assumption is the so-called limited treatment anticipation suggesting that there should be no effect of treatment before implementation (Roth et al., 2023).This assumption is usually satisfied whenever units (individuals or families) do not choose treatment status.The choice of residence is obviously not random, but a violation of this assumption would assume that families move in or out of the treatment area to seek or escape treatment.Whether this assumption holds on the individual level can be hard to investigate, but we generally have no information that suggests this to be a problem.A third assumption is also that treatment is absorbing, which implies that once an individual is treated, the individual will remain treated in the next period.This assumption will be satisfied if individuals who are treated do not move to another county after treatment, which is possible, but likely not a severe issue.We also exclude two of the bordering counties making individuals who move here fall out of the sample.
The distribution of the dependent variable, with a mean censored at zero, suggests that Poisson regression estimates should be included as well.Unfortunately, the software used here does not enable such estimates.Due to the applied outcome, the OLS is more vulnerable to heteroskedasticity with the dependent variable at hand (Woolridge, 2020); thus, we apply robust standard errors against this.Second, we risk negative predicted values in the outcome, but this is limited to <2% of the observations, which relax this issue (Woolridge, 2020, p. 577).Further, it should also be noticed that the interpretation of average treatment effects using non-linear estimators (e.g., logit, tobit) is not a straightforward exercise (Puhani, 2012).

Model specifications and results
The regression models apply the event-study design described in Section 3.2.The covariates included here are the leads measuring parallel pre-trends, while the lags estimate the treatment effect of introducing increased support for foster care families in the county on the number of moving across municipalities for children in fostercare.The models also include control variables for the number of individuals in a household (household size), weighted household income in 2015 NOK and age.As suggested by Wooldridge (Woolridge, 2020, p. 468), we ran both a pooled OLS model (Equation ( 1)) with first difference estimates as well as OLS with fixed effects estimates (Equation ( 2)), which includes unit fixed effects to control for time-invariant unobserved heterogeneity.Both models use municipal and time-fixed effects.The standard errors in both models are clustered against autocorrelation and heteroskedasticity on municipalities responsible for the placement of the children.This is because children are clustered into protective services that are administered by municipalities, and the municipality responsible for the children can vary from the registered municipality of the residence.
The dependent variable is the number of times a child moved within a year.This ranged from 0 up to 4 times.Here, λ t is the time-fixed effects, while α i is the individual fixed effect in the model ( 2).P ℓ μ ℓ are the coefficients from all the leads and lag variables at relative times ℓ.E i is the time of treatment for unit i. X i,t is the additional control variables, while u i,t is the error term.
Leads and lags are applied as a measurement for an individual i being E periods away from treatment at time t.It thus works as a "time until/since treatment" variable that measures the years until (leads) or after (lags) an individual received treatment.The coefficient from the leads and lags are typically interpreted as measuring the effect of treatment at different lengths of exposure to the treatment (Callaway & Sant'Anna, 2021, p. 208).Table 3 displays the distribution of individuals in different leads and lags.In the models lag 0 (treatment time) is omitted as a reference category as suggested by Cunningham (2021).
Table 4 and Figures 1 and 2 show that the two models yield very similar results.Neither model has significant differences between the treated and the untreated samples before treatment, thus indicating that the parallel trends assumption are satisfied.The pooled estimates show a significantly negative effect on moving from treatment in all 3 years after treatment.This effect is limited to the 2 years after treatment in the FE estimates.The findings thus suggest a significant but short-term effect from increased support to foster care homes.
The coefficient from lags in both models suggests that the frequency of moving was reduced by 0.4, which is a relative reduction of roughly 50%.The effect diminishes in the FE estimates, which potentially can be explained by the reform of 2022 and anticipatory measures in other counties and municipalities close to this.The results thus indicate a significant but short-term effect from the initiated measures.

DISCUSSION
The main finding from this study reveals a short-term decrease in terms of the number of moves in child welfare services after the implementation of measures to prepare for the upcoming child welfare services reform in Trøndelag county.These measures include "the incentive package," "BL," and "the care package."Although these measures had different approaches, their shared objective was to reduce the number of moves in OOHC by increasing different forms of support to foster families (Paulsen & Ytreland, 2022; The Norwegian Association of Local and Regional Authorities, 2021).The issue of placement instability in OOHC is a complex matter influenced by several factors such as the biological family, the youth, the foster family, and child welfare services (Backe-Hansen et al., 2019;Paulsen & Ytreland, 2022).Each of these actors is thus represented in measures aimed at stabilizing and securing OOHC.The results suggest that the combined effect of the implemented measures led to a substantial decrease in the number of moves for youth after the first measure was implemented in 2018.Note: Table 3 shows the number of treated and untreated individuals in each year.
T A B L E 4 Regression models.The most substantial decrease in terms of the number of moves occurred in the year after the implementation of "the incentive package" with a reduction from 0.09 to 0.05 moves on average in the treated area.The decline can thus be understood as an outcome of the measures in this package."The incentive package" increased stability and safety in OOHC primarily through increased supervision and resources directly to the foster families.The package further aimed at directing resources to foster homes considered at higher risk for placement instability where the initial period of placement was regarded as critical with a need for extra support (Paulsen & Ytreland, 2022).Providing increased resources directly to the families during this critical period may increase the feeling of security and stability for both the families and the youths in care.In turn, this can ease the adaptation to their new circumstances.Foster care is a complex relationship with many actors involved, and direct support and guidance from experienced and competent resources are thus likely to help establish successful relationships between all actors (Backe-Hansen et al., 2019; The Norwegian Association of Local and Regional Authorities, 2021).Furthermore, the results of this study indicate that giving additional support to families considered at risk for placement instability is effective.Such support may decrease the accumulation of risk factors that lead to placement instability, thus counteracting the trend of undesired moves.
Child welfare services in Trøndelag were given additional resources through a project called "barneløftet" ("BL") in terms of sharing resources between municipalities, increased budgets, and increased number of personnel.Among the measures were five additional counselors across the region to strengthen supervision (2021; The Norwegian Association of Local and Regional Authorities, 2021).The provision of additional resources can stimulate more cooperation between different municipalities that can help compensate for inequalities between the municipalities, particularly in a county with large differences in the number of citizens and distances such as Trøndelag and other Norwegian counties.Due to its geographical and demographic characteristics, there will be variation in available resources and the amount of experience within the municipalities (Backe-Hansen et al., 2014, 2019).This can have direct effects on the support provided to the families in terms of which type of support they obtain and how fast they can receive such support.Such differences pose a challenge in Norway where there is a principle of generalist local authority system meaning that all municipalities have the same responsibilities in providing public services for their residents (Norwegian Ministry of Local Government and Modernization, n.d.).
The results of this study indicate that the impact of "BL" on the number of moves is limited.This is indicated by no notable changes in the number of moves following the implementation of "BL" in 2020.However, the observed decrease after the implementation of "the incentive package" remained consistently low.These results indicate that the measures aimed at strengthening child welfare services did not have the desired effect.This is further emphasized by the evaluation of "BL," which concluded that even though many were content with the increased support, concerns were expressed regarding the accessibility of the measures from the incentive.The evaluation further indicated that there were still considerably different levels of support provided to different units of child welfare services received from "the children's promise" (2021).However, several incentives to strengthen child welfare services were implemented simultaneously in the region, and thus it is difficult to draw definite conclusions about the effect of each measure.
The incentives implemented in Trøndelag county impacted both families and child welfare services.The relative decrease of 50% in terms of the number of moves for the youths will have great impacts on the youths.A reduction in the number of placements allows youths to stay in one place over time and get the necessary stability to develop a sense of belonging in a social and spatial context.Facilitating placement stability and belonging increases the likelihood of feeling secure and included in a community.Security and belonging can facilitate development that affects outcomes later in life.The outcomes may include the completion of education, active participation in the labor market, being able to create meaningful relationships with others, and maintaining good levels of physical and mental health (Allen et al., 2021;Kääriälä, 2020;Konijn et al., 2019;Mendes & Moslehuddin, 2004).
"To move" can have different meanings and can be measured in a number of ways.Moving is defined in this article as a registered move between municipalities primarily because of the characteristics of the available data.This definition has some limitations.First, youths may have more placements than those officially registered, for example, because of emergency placements between more permanent placements.As a result, there may be trends in placement patterns for youth in foster care in Trøndelag county, which is not possible to explore with the available data.Second, youth in foster care may also move between different homes within the same municipality.These types of moves might have different implications for the individuals than moving between municipalities because they may remain part of the same community, for example, related to social groups and schools.Moreover, moving within a municipality might allow the youth to keep the same contact person within child welfare services and thus experience stability in care even when relocating homes.Inter-municipality moves are likely occurring more frequently in bigger municipalities, which also tend to be urban.It is further expected that there are differences in the provision of child welfare services in urban and rural areas in terms of resources and experience (Backe-Hansen et al., 2014, 2019;Kojan & Storhaug, 2021).
The results of this study indicate desired outcomes of the measures implemented to prepare Trøndelag for the upcoming child welfare services reform.The most notable reduction in the number of moves occurred after the implementation of "the incentive package," thus indicating its significant impact.This measure was intended to be a temporary incentive and was completed in 2020.The question remains as to whether the desired results from this package will continue after the incentive is completed, a concern also raised by Paulsen and Ytreland (2022).It is unclear if the practices and experiences gained through the package will continue to contribute to positive effects or if the desired effects are a direct result of increased resources.Assessing the continued effect of the measures described here will be more difficult because there will no longer be a natural comparison group that does not receive treatment after the child welfare services reform is implemented.Exploring movement patterns after the reform is initiated will be interesting due to the structural changes in OOHC and the potential for disparities in services across the country.Municipalities were responsible for a minimum of four follow-up sessions per year with foster families before their responsibility was increased with the new child welfare reform, and great inequalities have been detected in this type of support (Backe-Hansen et al., 2019;Paulsen & Ytreland, 2022;Pedersen et al., 2022).Questions are thus raised as to whether all municipalities can provide the support that foster families legally require, which is important for increased placement stability for youths in OOHC (Pedersen et al., 2022;Vanderwill et al., 2021).

CONCLUSION
This study employs an event-study design with difference-in-difference estimates to examine how the number of moves in out-of-home care was affected by incentives implemented in Trøndelag county before the child welfare services reform in 2022.The background for this study is that children in OOHC move three times as frequently as other children.This is problematic because of the negative implications that moving is known to have on youth development and future outcomes (Kääriälä, 2020;Paulsen et al., 2022;Paulsen & Ytreland, 2022)."The incentive package" and "BL" were implemented in the years leading up to the reform to increase stability and safety to all actors in OOHC through increased resources in terms of supervision, personnel, and finances.Both incentives aimed at reducing the number of moves for youths in child welfare services (Paulsen & Ytreland, 2022; The Norwegian Association of Local and Regional Authorities, 2021).
Our findings reveal that the implemented measures reduced the number of moves.The most significant impact is seen after the implementation of "the incentive package" where the number of moves decreased from 0.09 to 0.04 after initial treatment, which is close to the rest of the youth population.Although the number of moves increased again over time, it was consistently lower than before treatment in the following years, no significant decrease was observed after the implementation of "BL."These findings suggest that increased support directly to foster care homes do have a significant effect on reducing the number of moves in OOHC.
The main limitations of this study are linked to the characteristics of the data material.We only observe 85% of children in OOHC-these are those under the care of the child welfare administration.Further, we are not able to make distinctions between planned and unplanned moves in our analysis or indicate whether the registered moves include change of caregivers.These are interesting distinctions for future research.We can only track youths with registered moves across municipality boundaries meaning that unregistered moves and moves within municipalities are not included in this study.Thus, we cannot indicate if the incentives affected such moves.Although, we have indications that cross-municipal moves are likely to cover a significant share of moves for children in OOHC.Importantly, we do not find moving patterns to be different between the treated and the control group for the baseline population.This issue was further addressed in the analysis by controlling for parallel trends in baseline outcomes before the reform.Additionally, one could expect within-municipality moves to happen more frequently in urban municipalities because of their size and demographic, which is another aspect that we cannot unpack with the available data.

DATA AVAILABILITY STATEMENT
The data used is available through the interface Microdata.no.The code used for the analysis can be shared by request to the authors.
Norunn Hornset https://orcid.org/0009-0006-9654-9031BårdSmedsvik https://orcid.org/0000-0001-8843-0256APPENDIXT A B L E A 2 Descriptive statistics on age.The table shows average age for children in foster care and the average age at placement in the period.Descriptive statistics on gender.The table shows the % of females in Norwegian foster care, in total, treated, and not-treated areas.T A B L E A 1 Descriptive statistics of additional covariates.Standard deviation in parentheses.Income in 2015 NOK (thousands).
Note:T A B L E A 3 Note:Note: